Openberth, developed by Amirsoleimani, is a self-hosted deployment platform built primarily with Go. It enables developers, teams, and individuals to manage applications on their own infrastructure. This mcp category program distributes its server component via an installer script, setting up essential dependencies like Docker and gVisor. The CLI is provided as downloadable static Go binaries, designed for automatic OS and architecture detection. Openberth focuses on deploying various application types within sandboxed gVisor containers, offering features that cater to traditional CLI workflows and advanced AI client integrations.

What does Openberth connect to?

Openberth operates as a self-hosted deployment platform, connecting to and managing sandboxed gVisor containers. These containers are used for deploying applications written in Node.js, Python, and Go, alongside static sites. It also exposes an MCP server endpoint, facilitating integration with various AI clients. This design allows for secure and isolated execution environments for hosted applications while providing a dedicated interface for AI-driven operations. The platform's server setup involves Docker, gVisor, Caddy, SQLite, and systemd.

Core features and tools

Openberth provides a set of callable tools for managing deployments. These include berth_deploy for application deployment, berth_update for pushing changes, and berth_status to check application health. Users can view logs with berth_logs, list deployments using berth_list, and manage access with berth_protect. The platform also offers berth_destroy to remove applications and additional sandbox tools. It supports HTTP API, MCP server, and stdin pipe as transport protocols, with the MCP endpoint specifically using Streamable HTTP.

Who is it for?

Openberth is designed for developers, teams, and individuals who require a self-hosted deployment platform. It is particularly valuable for those interested in integrating AI assistants into their code deployment and management workflows. The platform supports traditional CLI deployment of various application types, including React, Python, Go, and Jupyter notebooks. Its secure secret management and AI integration features appeal to those building modern, AI-assisted development environments. Skip this if you do not plan to host your own deployment infrastructure.

How to get started

The server authenticates using per-user API keys, which can be managed via "berth-admin user add" or through OpenID Connect for login. The MCP server endpoint specifically uses Streamable HTTP transport.

Use cases

• AI-assisted deployment, where an AI writes and deploys code, referencing encrypted secrets by name to obtain a live HTTPS URL.
• Traditional CLI deployment of various application types, such as React, Python, Go, and Jupyter notebooks, using the "berth deploy" command.
• AI-driven live iteration and debugging, allowing AI to push updates with hot reload and pull logs directly within a chat interface.

Openberth offers a self-hosted platform for deploying applications within sandboxed gVisor containers. Its support for AI client integration, including Claude.ai and Cursor, provides distinct workflows for modern development. The platform handles secret management and offers per-user API key authentication with OIDC support. Developers seeking control over their deployment environment while exploring AI-assisted operations will find Openberth useful.